A fan shaft torsional vibration suppression method based on robust nonlinear adaptive control

Abstract

The application belongs to the field of control and energy, and specifically discloses a wind turbine shaft torsional vibration suppression method based on robust nonlinear adaptive control, which comprises the following steps: firstly, modeling the shaft model and the electrical model of the double-fed wind turbine, analyzing the shaft torsional vibration initiation mechanism, and defining the wind turbine shaft torsion angle deviation; then, converting the double-fed wind turbine model into a second-order nonlinear system form, designing a shaft torsional vibration suppressor for the converted model, ensuring the asymptotic tracking performance of the converted system, and effectively suppressing the shaft torsional vibration problem of the wind turbine when it encounters a large disturbance. In the designed shaft torsional vibration suppressor, the uncertain dynamics in the wind power grid-connected system are approximated by an online neural network, and the external disturbance and the reconstruction error of the neural network are compensated by an error sign integral robust term. The application can overcome the uncertainty in the existing model, enhance the robustness of the control system, and has anti-interference ability.

Description

Technical Field

[0001] This invention belongs to the field of control and energy, and particularly relates to a method for suppressing torsional vibration of wind turbine shaft systems based on robust nonlinear adaptive control. Background Technology

[0002] Over the past 20 years, the continuous increase in the scale of wind power grid connection has brought enormous challenges to the stable operation of the power system, among which the shaft torsional vibration problem has attracted much attention from scholars. When the mechanical or electromagnetic power of the wind turbine experiences large disturbances, the torsional vibration phenomenon of the wind turbine shaft system becomes particularly severe, leading to serious mechanical fatigue. This not only causes serious economic losses but may also create hidden dangers for grid stability, because the strong shaft torsional vibration of the wind turbine may trigger or aggravate other oscillation behaviors in the grid, thus seriously threatening the stable operation of the power system. Therefore, it is necessary to design effective torsional vibration suppressors to address the problem of wind turbine shaft torsional vibration under large disturbances.

[0003] However, existing technologies typically analyze and control torsional vibration of grid-connected wind turbine shafts based on linear system theory. The proposed torsional vibration suppressors for wind turbine shafts only show good performance for approximately linearized systems near the equilibrium point. Furthermore, another nonlinear wind turbine shaft torsional vibration suppression method based on sliding mode control has been proposed. However, in this nonlinear method, when the grid-connected system encounters large disturbances such as sudden wind speed changes, grid faults, or wind power frequency regulation, the approximately linearized system at the equilibrium point will deviate significantly from the actual system, greatly reducing the shaft suppression performance of the designed torsional vibration suppressor, and potentially even causing it to completely lose its torsional vibration suppression function. Therefore, a nonlinear wind turbine shaft torsional vibration suppressor based on sliding mode control (SMC) has been proposed. Sliding mode control has the ability to handle the nonlinear characteristics of the system and can effectively suppress torsional vibration of the wind turbine shaft after the system encounters large disturbances, but it may exhibit chattering phenomena caused by its discontinuous behavior. In fact, nonlinear adaptive control theory is a more effective and mature method than SMC in dealing with the nonlinear characteristics of a system. Therefore, the limitations of existing methods can be eliminated by using nonlinear adaptive control. Summary of the Invention

[0004] Due to the interaction and coupling between the wind turbine and the grid, the "mechanical friendliness" of a "grid-friendly" wind turbine is difficult to guarantee under complex grid conditions. On the one hand, grid disturbances / faults can cause disturbances in the electromagnetic torque of the generator, increasing the load on the wind turbine. On the other hand, when there are large disturbances in wind speed, the start-up of the turbine, or the sudden shutdown of the turbine, the mechanical power of the turbine will change abruptly. The inertia of the turbine shaft system is insufficient to significantly mitigate such large disturbances. Both of these torque changes will cause the turbine to face serious shaft torsional vibration problems. If effective suppression measures are not taken, the torsional vibration of the turbine shaft system will not only seriously affect the life of the turbine and cause high economic losses, but may also worsen other oscillations in the grid, thus generating a series of complex power system stability problems. Therefore, conducting research on shaft damping control of wind turbines under large disturbances has become an important challenge in the stability control of wind power grid-connected systems.

[0005] Uncertain dynamics of the grid-connected system can be approximated using an online neural network, and compensated online using an adaptive control law. To mitigate the impact of external time-varying disturbances and neural network reconstruction errors, a robust error sign integral term can be used for compensation, ensuring the continuity of the obtained control signal and avoiding actuator chattering. The key to this control method lies in designing a new control law that, under conditions of uncertain dynamics and external disturbances in the wind power grid-connected system, achieves asymptotic tracking performance while maintaining control signal continuity, effectively avoiding control signal chattering and further improving control performance and the robustness of the shaft torsional vibration suppressor.

[0006] Considering the torsional vibration problem of wind turbine shaft systems under large disturbances, this invention proposes a continuous control strategy combining neural networks and error sign integral robustness. This invention does not require prior knowledge of uncertainties, external disturbances, and neural network reconstruction errors in wind power grid-connected systems. The uncertain dynamics of the system are approximated by an online neural network, while external time-varying disturbances and neural network reconstruction errors are compensated by an error sign integral robustness term. This invention not only effectively handles the torsional vibration problem of wind turbine shaft systems under large disturbances but also exhibits good robustness in handling model parameters and wind power uncertainties. The specific implementation of the method of this invention includes the following steps:

[0007] The objective of this invention is achieved through the following technical solution:

[0008] A method for suppressing torsional vibration of a wind turbine shaft system based on robust nonlinear adaptive control, comprising the following steps:

[0009] Step 1: First, model the doubly-fed wind turbine and provide its shaft system model:

[0010]

[0011]

[0012]

[0013] Where w wt and w g Indicates the shaft rotation speeds on the wind turbine side and the generator side, respectively; T m and T e Representing mechanical torque and electromagnetic torque respectively; θ sh Indicates the shaft torsion angle; H wt and H g K represents the inertia of the generator and the wind turbine, respectively. sh and D sh These are the damping coefficient and stiffness coefficient of the fan shaft system, respectively; w B A reference value indicating the rotational speed of the fan shaft system;

[0014] Step 2, select the output variable y as the wind turbine shaft twist angle θ. sh Its equilibrium point θ sh0 The deviation, i.e., y(t) = Δθ sh (t)=θ sh (t)-θ sh0 t represents time; electromagnetic torque compensation amount T e,comp To control the input u, the reference value of the doubly-fed induction generator's electromagnetic torque is equal to the electromagnetic torque T under the maximum power point tracking strategy. e,MPPT The sum of the electromagnetic torque compensation amount;

[0015] Step 3: Calculate the second derivative of the deviation angle, while considering the existence of external disturbances, and obtain the following expression:

[0016]

[0017] In the formula, g(Y) and f(Y) are complex nonlinear functions with respect to the output variable y and its derivative, and v(t) is a time-varying disturbance;

[0018] Step 4, for Perform order reduction and define the filtered output vector. Where λ is a positive constant, and then the first derivative with respect to r(t) is expressed as:

[0019]

[0020] In the formula, F(Y)=f(Y)+λw B (w wt -w g );

[0021] Step 5, Define Where a is a positive constant, and let T(Y) = g -1 (Y), obviously T(Y)>0, then for Taking the derivative and multiplying it by T(Y), we obtain the following expression:

[0022]

[0023] In the formula, Z = [Y T u T ] T H(Z) and V(t) are auxiliary functions that include the unknown dynamics of the wind power grid-connected system;

[0024] Step 6: Considering that H(Z) does not contain time-varying disturbance v(t), it can be approximated by a neural network, i.e.:

[0025] H(Z)=W T Φ(Z)+ε(Z)

[0026] In the formula, Φ(Z) is the activation function, ε(Z) is the reconstruction error, and W is the weight matrix. We can let... For an estimate of W, the approximate expression of H(Z) is: Here, W and ε(Z) satisfy the boundedness property, that is, ||W||≤W m ,||ε(Z)||≤ε m0 , ε m0 and W m Represents an unknown positive integer;

[0027] Step 7, Design the control law for the shaft torsional vibration suppressor:

[0028]

[0029] Where k s Let γ(t) be a positive constant and γ(t) be an adaptive variable. Further substituting the above equation into... From the expression:

[0030]

[0031] in,

[0032] Step 8, Estimation of the output layer weight matrix in the control law The update law for the adaptive parameter γ(t) can be designed as follows:

[0033]

[0034]

[0035] Among them l w ,σw It is a positive number;

[0036] Step 9: Use the output u of the designed shaft torsional vibration suppressor as the electromagnetic torque compensation amount, and then change the required electromagnetic torque reference value for torque control from the original maximum power tracking torque T. e,MPPT In addition, an electromagnetic torque compensation is added to achieve the effect of suppressing shaft torsional vibration.

[0037] Furthermore, in step 3, the expressions for g(Y) and f(Y) are as follows:

[0038]

[0039] Furthermore, in step 5, the unknown dynamics in the wind power grid-connected system are divided into two parts, represented by auxiliary functions H(Z) and V(t), whose expressions are as follows:

[0040]

[0041]

[0042] Furthermore, in step 7, the uncertain term Their derivatives are bounded, i.e., they satisfy:

[0043]

[0044]

[0045] in, All are unknown positive numbers.

[0046] Furthermore, in steps 7 and 8, the Lyapunov function is selected as V. n , is represented as:

[0047]

[0048] In the formula,

[0049] It is an unknown positive number. and functions

[0050] This can prove all signals of the controlled system It satisfies boundedness.

[0051] The beneficial effects of this invention are as follows: This invention, based on a robust nonlinear adaptive control method for suppressing torsional vibration of wind turbine shafts, achieves the goal of suppressing torsional vibration of wind turbine shafts under large disturbances encountered by the wind power grid-connected system. Simultaneously, it improves the robustness of the shaft torsional vibration suppressor in handling various uncertainties. This control method compensates for bounded reconstruction errors and external disturbances by introducing robust error sign integrals, and designs an online weight update algorithm that does not require knowledge of the actual neural network weight information, thereby ensuring the asymptotic tracking performance of the controlled system, and ensuring that all closed-loop system signals are bounded. This control method can achieve asymptotic tracking performance even under uncertain dynamics and external disturbances in the wind power grid-connected system. Furthermore, this control method overcomes uncertainties in existing models, enhances the robustness of the control system, and has anti-interference capabilities. In addition, the control signal designed in this method is continuous, effectively avoiding the chattering problem of the actuator, further ensuring the safe and stable operation of the wind power grid-connected system. Attached Figure Description

[0052] Figure 1 This is a flowchart of a method for suppressing torsional vibration of a wind turbine shaft system based on robust nonlinear adaptive control according to the present invention.

[0053] Figure 2 This is a block diagram of the control system of the present invention;

[0054] Figure 3 This is a diagram showing the torsional angle response curve of the fan shaft system under a three-phase short-circuit fault according to the present invention.

[0055] Figure 4 This is a graph showing the torsional angular velocity response of the fan shaft system under a three-phase short-circuit fault according to the present invention. Detailed Implementation

[0056] The present invention will be further described below with reference to the accompanying drawings and specific embodiments. The following examples are only used to more clearly illustrate the technical solutions of the present invention, and should not be used to limit the scope of protection of the present invention.

[0057] This invention provides a method for suppressing torsional vibration of a wind turbine shaft system based on robust nonlinear adaptive control. The flowchart of this control method is shown below. Figure 1 As shown, the control system block diagram is as follows: Figure 2 As shown, the rotational speed w of the wind turbine side shaft system wt and the rotational speed w of the generator side shaft system g The torsional deviation Δθ of the fan shaft system is obtained by subtracting and integrating. sh Shaft system torsional deviation Δθ shThe data is transmitted to a robust nonlinear torsional vibration suppressor. The control law, calculated by the shaft torsional vibration suppressor, outputs control input variables, which are then transmitted to the rotor-side converter as electromagnetic torque compensation, thereby controlling the oscillation of the shaft torsional angle. The specific implementation of this method includes the following steps:

[0058] Step 1: Assume the grid-connected wind farm consists of six doubly-fed induction generator (DFIG) turbines, each with a rated power of 710 kW, and these six turbines are modeled as a single integrated DFIG turbine. First, consider its shaft system model, represented as:

[0059]

[0060]

[0061]

[0062] Where w wt and w g Indicates the shaft rotation speeds on the wind turbine side and the generator side, respectively; T m and T e Representing mechanical torque and electromagnetic torque respectively; θ sh Indicates the shaft torsion angle; H wt and H g K represents the inertia of the generator and the wind turbine, respectively. sh and D sh These are the damping coefficient and stiffness coefficient of the fan shaft system, respectively; w B A reference value indicating the rotational speed of the fan shaft system;

[0063] Step 2, select the wind turbine torsion angle deviation Δθ sh =θ sh -θ sh0 For the output variable y, the electromagnetic torque compensation amount T e,comp To control the input u, which is output by the shaft torsional vibration suppressor, the reference value of the electromagnetic torque of the doubly-fed wind turbine is equal to the electromagnetic torque T under the maximum power point tracking (MPPT) strategy. e,MPPT The sum of the electromagnetic torque compensation amount;

[0064] Step 3: Calculate the second derivative of the deviation angle, while considering the existence of external disturbances, and obtain the following expression:

[0065]

[0066] In the formula, The expressions for g(Y) and f(Y) can be derived from the above shaft system model;

[0067] The expressions for g(Y) and f(Y) are as follows:

[0068]

[0069] Step 4: To facilitate the design of the shaft torsional vibration suppressor, the second-order equation above is reduced in order. Define the filtered output vector. Where λ = 8. Then, taking the first derivative with respect to r(t), the expression is:

[0070]

[0071] In the formula, F(Y)=f(Y)+λw B (w wt -w g ).

[0072] Step 5, Define Where a = 2. (Regarding...) Differentiate and multiply the derivative by T(Y) = g -1 (Y), let Z = [Y] T u T ] T The resulting expression is as follows:

[0073]

[0074] The unknown dynamics in the wind power grid-connected system are divided into two parts, represented by auxiliary functions H(Z) and V(t), and their expressions are as follows:

[0075]

[0076]

[0077] Step 6: Considering that the auxiliary function H(Z) does not contain the time-varying disturbance v(t), it can be approximated by a neural network, i.e.:

[0078] H(Z)=W T Φ(Z)+ε(Z).

[0079] Furthermore, in practice, the weight matrix W is unknown, so we can let For an estimate of W, the approximate expression of H(Z) is:

[0080] Step 7, design the following control law:

[0081]

[0082] Where k s Let k be a positive constant, γ(t) be an adaptive variable, and k be chosen. s=20. At this point, the control input u can be used as the electromagnetic torque compensation value to calculate the reference value of the doubly-fed wind turbine's electromagnetic torque. Inputting this value into the rotor-side converter will achieve the effect of torsional vibration suppression. Further substituting the above formula into... From the expression:

[0083]

[0084] in,

[0085] Among them, the uncertain terms Their derivatives are bounded, i.e., they satisfy:

[0086]

[0087]

[0088] in, All are unknown positive numbers.

[0089] Step 8, in addition, the estimated value of the output layer weight matrix in the control law. The update law for the adaptive parameter γ(t) can be designed as follows: Select l w =50, σ w =1e-6. The update law for γ(t) is designed as follows:

[0090]

[0091] Among them l w ,σ w It is a positive number;

[0092] Step 9: Use the output u of the designed shaft torsional vibration suppressor as the electromagnetic torque compensation amount, and then change the required electromagnetic torque reference value for torque control from the original maximum power tracking torque T. e,MPPT In addition, an electromagnetic torque compensation is added to achieve the effect of suppressing shaft torsional vibration.

[0093] Furthermore, in steps 7 and 8 above, a Lyapunov function is selected:

[0094]

[0095] In the formula,

[0096]

[0097] Differentiating the above equation, we get:

[0098]

[0099] This can prove all signals of the controlled system It satisfies boundedness.

[0100] The parameters of the doubly fed fan and the proposed fan shaft suppressor are shown in Table 1.

[0101] Table 1. Parameters of the doubly-fed motor and the proposed fan shaft suppressor

[0102]

[0103] Considering a large disturbance: at 13.5s, a three-phase short-circuit fault occurs on the bus, and the fault is cleared 0.2s later.

[0104] Figure 3-4 This is a graph showing the response curves of the difference in angular velocity and shaft torsion angle between the wind turbine and generator sides under a large disturbance scenario. From... Figure 3-4 Simulation results show that when a three-phase short-circuit fault occurs in the wind power grid-connected system, the proposed shaft torsional vibration suppressor can effectively estimate the nonlinear part of the system in real time, and at the same time, timely and dynamic compensation of the system disturbance further improves its performance and suppresses the torsional vibration of the wind turbine shaft at a relatively fast speed.

[0105] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for suppressing torsional vibration of a fan shaft system based on robust nonlinear adaptive control, characterized in that, The method comprises the following steps: Step 1, firstly, the doubly-fed wind turbine is modeled, and a shafting model is given: ， where w wt and w g represent the shaft speed of the wind turbine side and the generator side respectively; T m and T e represent the mechanical torque and the electromagnetic torque respectively; θ sh represents the shaft torsion angle; H wt and H g represent the inertia of the generator and the wind turbine respectively; K sh and D sh are the damping coefficient and the stiffness coefficient of the wind turbine shaft respectively; w B represents the reference value of the wind turbine shaft speed; Step 2, the output variable y is selected as the wind turbine shaft torsion angle θ sh deviation from its equilibrium point θ sh0 , i.e. t represents time; the electromagnetic torque compensation amount T e,comp is the control input u, then the electromagnetic torque reference value of the doubly-fed wind turbine is equal to the sum of the electromagnetic torque T e,MPPT at the maximum power point tracking strategy and the electromagnetic torque compensation amount. Step 3, the second derivative of the deviation angle is calculated, and the existence of disturbance is considered externally, and the following expression is obtained: ， where g(y) and f(y) are complex nonlinear functions of the output variable y and its derivatives, and v(t) is a time-varying disturbance; The expressions of g(Y) and f(Y) are as follows: ， ； Step 4, to reduce the order, define the filtered output vector where λ is a constant, and then take the first derivative of r(t), which is given by: ， In the formulae, ; Step 5, Definition where a is a positive constant and let T(Y) = g -1 (Y), it is clear that T(Y) > 0, and then taking the derivative and multiplying by T(Y) gives the expression: T(Y) = aY ， In the formula, H(Z) and V(t) are auxiliary functions containing unknown dynamics of the wind power grid-connected system; Wherein, the unknown dynamics in the wind power grid-connected system are divided into two parts, which are expressed by auxiliary functions H(Z) and V(t), and the expressions are as follows: ， ； Step 6, considering that the time-varying disturbance v(t) is not included in H(Z), therefore, it can be approximated by a neural network, that is: ， where Φ(Z) is the excitation function, ε(Z) is the reconstruction error, and W is the weight matrix, which can be allowed to be an estimate of W, then an approximate representation of H(Z) is where W and ε(Z) satisfy boundedness, i.e., there exist positive constants C and D such that , , ε m0 and W m represent unknown constants; Step 7, the control law of the shafting torsional vibration suppressor is designed: ， where k s is a positive constant, and γ(t) is an adaptive variable. Substituting the above equation into The expression is obtained. ， wherein , , ; Step 8, the output layer weight matrix estimation value in the control law The update law of the adaptive parameter γ(t) can be designed as: ， ， wherein l w ,σ w is a normal number; Step 9, the output u of the designed shafting torsional vibration damper is used as an electromagnetic torque compensation amount, and then the required electromagnetic torque reference value of the torque control is compensated by the original maximum power tracking torque T e,MPPT , and an electromagnetic torque compensation amount is additionally added, thereby achieving the effect of suppressing shafting torsional vibration.

2. The robust nonlinear adaptive control based fan shaft torsional vibration suppression method according to claim 1, wherein, In step 7, the uncertain term , and their derivatives are bounded, i.e. satisfy: ， ， wherein are unknown constants.

3. The robust nonlinear adaptive control based fan shaft torsional vibration suppression method of claim 2, wherein, In the steps 7 and 8, Lyapunov function is selected as V n , which is expressed as: ， In the formulae, , , , , and function i.e. all signals of the controlled system can be proven to satisfy boundedness.

Images